Skip to content

[mlir][shard] Hardening sharding propagation and partitioning #183028

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
fschlimb merged 2 commits into from
Feb 24, 2026
Merged

[mlir][shard] Hardening sharding propagation and partitioning #183028

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
739beec
hardening sharding propagation and partitioning
fschlimb Feb 24, 2026
e90b7c3
adresed review comment
fschlimb Feb 24, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions mlir/lib/Dialect/Shard/Transforms/Partition.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -532,7 +532,8 @@ shardedBlockArgumentTypes(Block &block,
block.getArguments(), std::back_inserter(res),
[&symbolTableCollection](BlockArgument arg) {
auto rankedTensorArg = dyn_cast<TypedValue<RankedTensorType>>(arg);
if (!rankedTensorArg || rankedTensorArg.getType().getRank() == 0) {
if (!rankedTensorArg || rankedTensorArg.getType().getRank() == 0 ||
rankedTensorArg.use_empty()) {
return arg.getType();
}

Expand Down Expand Up @@ -666,14 +667,16 @@ partitionOperation(ShardOp shardOp, IRMapping &partitionMap,
static LogicalResult checkFullyAnnotated(Block &block) {
for (const BlockArgument &arg : block.getArguments()) {
auto rankedTensorArg = dyn_cast<TypedValue<RankedTensorType>>(arg);
if (!rankedTensorArg || rankedTensorArg.getType().getRank() == 0)
if (!rankedTensorArg || rankedTensorArg.getType().getRank() == 0 ||
rankedTensorArg.use_empty())
continue;

if (rankedTensorArg.getNumUses() > 1)
return emitError(block.getParent()->getLoc())
<< "Cannot partition: expected a single use for block argument "
<< arg.getArgNumber() << " in block "
<< block.computeBlockNumber();

Operation *useOp = *rankedTensorArg.getUsers().begin();
auto shardOp = dyn_cast<ShardOp>(useOp);
if (!shardOp)
Expand Down
4 changes: 4 additions & 0 deletions mlir/lib/Dialect/Shard/Transforms/ShardingPropagation.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -379,6 +379,10 @@ struct ShardingPropagation
shardingOp.printLoopTypesAndIndexingMaps(llvm::dbgs());
});

// Nothing to propagate if there is no sharding annotation in the block.
if (block.getOps<shard::ShardOp>().empty())
return;

auto traverse = [&](auto &&range, OpBuilder &builder,
const char *order) -> bool {
for (Operation &op : range) {
Expand Down
Loading